TATA-binding Protein Activates Transcription ... - Semantic Scholar

4 downloads 0 Views 6MB Size Report
Christopher J. BrandlS, Joseph A. Martens#, Patricia C.-Y. Liaw, Angela M. Furlanetto, and ...... Giniger, E., Varnum, S. M., and Ptashne, M. (1985) Cell 40, 767-774 ... Simon, M. C., Fisch, T. M., Benecke, J. B., Nevins, J. R., and Heintz, N. (1988).
Vol. 267, No. 29, Issue of October 15, pp. 20943-20952,1992 Printed in U.S.A.

THEJOURNAL OF BIOLOGICAL CHEMISTRY 0 1992 by The American Society for Biochemistry and Molecular Biology, Inc

TATA-binding Protein Activates Transcription When Upstream of a GCN4-binding Site in a Novel Yeast Promoter* (Received for publication, April 28, 1992)

Christopher J. BrandlS, Joseph A. Martens#, PatriciaC.-Y. Liaw, Angela M. Furlanetto, and C. Richard Wobben From the Department of Biochemistry, University of Western Ontario, London, Ontario N6A 5C1, Canada and the qMerck Research Laboratories, Rahway, New Jersey 07065

In the gal-his3 hybrid promoter, his3-GG1, GCN4 stimulates transcription at the position normally occupied by a TATA element. This expression requires two elements within gall-10 sequences, a REB1-binding site and a second element, Z, which resides 20 base pairs upstream of the GCN4-binding site. No obvious TATA element is present in this promoter. To characterize the function of Z, we replaced it with short random oligonucleotidesand selected for expression in vivo. Fourteen elements were identified and classified into groups based upon sequence and phenotypic similarities. Group 1elements contained functional TATA sequences that were essential for activity. TATA elements can thus function when positioned upstream of a GCN4-binding site. The Group 2 elements activated transcription poorly when usedas conventional TATA elements; however, mutational analyses demonstrated that theiractivityrequired TATA-like sequences. These TATA-like sequences bound the yeast TATAbinding protein (TBP) poorly in vitro but function in vivo as TBP interaction sitesbased upon two criteria. First mutations that improved their TATA character correspondingly improved function and second their activity could be enhanced in thepresence of an altered binding specificity mutant of TBP. Furthermore, the Group 2 elements enabled the identification of mutations outside of the TATA-like core that contribute to transcriptional activation without adversely affecting TBP binding. The finding that low affinity TBP-binding sites canbe used at unconventional positions suggests that many “TATA-less” promoters contain a cryptic interaction site for TBP.

As the underlying basis for the correct assembly of the transcriptional machinery, promoter structure plays a significant role in the mechanism of transcriptionalactivation. This is the case for most eucaryotic promoters transcribed by RNA polymerase I1 that contain a TATA element,consensus sequence TATAAA, proximal to the start site of transcription. The TATA element is important for both the rate andaccuracy of transcription in vivo and in vitro (for reviews, see Guarente, 1988; Struhl, 1989). TATA elements function by

* This work was supported by funds from the Medical Research Council of Canada, the Academic Development Fund of the University of Western Ontario, and the London Life Insurance Co. The costs of publication of this article were defrayed in part by the payment of page charges. This article musttherefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. $ Recipient of a Medical Research Council scholarship. To whom correspondence should be addressed. Tel.: 519-661-3908 Fax: 519661-3175. § Recipient of a Medical Research Council studentship.

recruiting TFIID (Davison et al., 1983; Parker and Topol, 1984; Nakajima et al., 1988), aprotein complex generally believed to be a component of the basic transcriptional machinery that includes RNA polymerase I1 as well as a variety of other factors including TFIIA, TFIIB,TFIIE,TFIIF, TFIIH, and TFIIJ(for review, see Stone et al., 1991). In vitro the binding of TFIID is the first step in transcriptioncomplex assembly which in turn facilitates the recruitment of TFIIB and ultimately the other basic factors to the promoter (Fire et al., 1984; Reinberg et al., 1987; Van Dyke et al., 1988; Buratowski et al., 1989; Maldonado et al., 1990). TFIID may also promote a nucleosomal structure on the template DNA which is competent for transcriptional initiation. Prior binding of TFIID will relieve transcriptional repression caused by nucleosomal assembly in vitro (Workman and Roeder, 1987). TFIID has been defined by biochemical fractionation and is composed of multiple subunits in higher eucaryotes (Dynlacht et al., 1991; Meisterernst et al., 1991; Pugh and Tjian, 1991; Safer et al., 1991; Tanese et al., 1991; Timmers and Sharp, 1991),the principal component of which isthe TATAbinding protein (TBP)’ (Berger et al., 1990; Buratowski et al., 1988; Cavallini et al., 1988; Kelleher et al., 1990; Pugh and Tjian, 1990). The exact composition of the TFIID complex may in fact determine its functional specificity (for review, see Sharp, 1992). In yeast TBP fractionates as a monomeric protein (Buratowski et al., 1988); however, the requirement for additional factors for activated transcription (Berger et al., 1990; Kelleher et al., 1990; Flanagan et al., 1991) and the recent finding that TBP is required for RNA polymerase I and I11 transcription (Cormack and Struhl, 1992; Schultz et al., 1992) suggest that less tightly associated factors may also allow promoter-specific function of yeast TBP. To facilitate the discussion we shall use TFIID toindicate such complexes. Despite the significance of TFIID in theinitiation process, many eucaryotic promoters lack sequences which resemble functional TATA elements (for examples see Carcamo et al., 1990; Smale and Baltimore, 1989 and references therein). In addition, different transcriptional activator proteins function in combination with various TATA element sequences with different efficiencies (Struhl, 1986;Gill and Ptashne, 1988; Homa et al., 1988; Simon et al., 1988; Harbury and Struhl, 1990; Wefald et al., 1990; Taylor and Kingston, 1990). These findings raise three questions. Is TFIID required for transcription of allpromoters? If so how might it be recruited to TATA-less promoters and can other factors substitute for TFIID? Previously, we have shown that theyeast activator protein GCN4, which normally activates transcription when bound upstream of a TATA element, can activate transcription when The abbreviations used are: TBP, TATA-binding protein; AT, 3amino-1,2,4-triazole; PMSF, phenylmethanesulfonyl fluoride; IPTG, isopropylthio-8-galactoside;DTT, dithiothreitol; bp, base pair(s).

20943

20944

Promoter-specific TATA Sequences

its binding site replaces the required TATA elementof a gall10/his3 hybrid promoter (Brandl and Struhl,1990; Chen and Struhl, 1989). Unlike transcription of equivalent gall-l0/his3 hybrid promoters that require a TATA element and are responsive to GAL4-mediated induction(Chen andStruhl, 1988), transcription of the chimeric promoter his3-GG1 (and its truncatedderivative his3-GG155, Fig. 1)occurs in glucose and is responsive to GCN4 induction. This promoter thus serves as a model system for regulated TATA-less promoters. Transcription of the his3-GG promoters requires three elements found upstream of the start sites: the GCN4-binding site centered at -41 relative to the +1 site of initiation, and two elements Q and Z found within the gall-10 sequences (Brandl and Struhl, 1990). Q overlaps the second most proximal GAL4-binding site, approximately 130 base pairs (bp) upstream from the GCN4-binding site. Q element function in vivo correlates directly with the ability of the element to bind the protein REBl (QBP, GRFZ) (Brandl and Struhl, 1990). REBl was first identified by its binding of an element within the 35 S ribosomal genes (Morrow et al., 1989) but has since been identified as a component of a diverse number of genetic elements including centromeres and telomeres (Chasman et al., 1990; Finley and West, 1989; Wang et al., 1990). Interestingly, the REB1-binding site in gall-10 forms a constraint to nucleosome positioning (Fedor et al., 1988), thus raising the possibility that transcriptional activation mediated by REBl occurs via alterations to nucleosome structure. The element Z activates transcription of his3-GG1 approximately 4-fold and was initially defined by deletion and disruption analysis to be approximately 20 bp upstream from the GCN4-binding site (Brandl and Struhl, 1990). More recently with directed mutations in this region’ we have mapped the element to gall-10 sequences from 315 to 329 (Johnston and Davis, 1984). There is no obvious TATA elementin this region. In fact, the replacement of Z with a consensus TATA element, with no consideration of flanking sequences, did not restore function (Brandl and Struhl, 1990). The gall-IO region contains multiple functional elementswhich could confer Z function in the context of the his3-GG promoter. In particular, Finley et al. (1990) have identified an element termed 0 4 which overlaps Z and is one of six negative control elements that repress transcription of gall-10 in the absence of galactose. To further characterize the Z element with the goals of understanding its roles in the transcription of his3-GG1 and the mechanism by which GCN4 can activate transcription in the absence of a TATA element, we have used a random selection approach to identify sequences that canactivate transcription in conjunction with GCN4 and REBl at the position occupied by Z. This approach allows for the identification of the range of elements that will provide a particular function without bias toward the native sequence. Surprisingly, using this selection procedure we now show that certain TATA elements can substitute for Z and thus can activate transcription when placed upstream of an activator protein. Moreover, promoter structure plays a significant role in determining the sequence requirements of the core TATA element andits flanking sequences. The his3-GG promoter configuration can tolerate a lower affinity binding of TBP than conventional promoter arrangements but has an additional requirement for specific sequences outside the TATA element core. These results support the view that TBP is required for the transcription of TATA-less promoters, with TBP recruited to a cryptic binding site that is functional in the context of a specific promoter structure.

* A. M. Furlanetto and C. J. Brandl, unpublished data.

MATERIALS ANDMETHODS

Construction of DNA Molecules-DNA manipulations were carried out using standard cloning techniques (Sambrook et al., 1989). Molecules were verified by restriction and DNA sequence analysis. All of the his3 alleles used contain a 6.1-kilobase fragment of yeast chromosomal DNA containingpet56, his3, and dedl genes (Struhl, 1985). These alleles were cloned into the u r d containing integrating vector, YIp55 (Struhl, 1986). Single strand oligonucleotideswere prepared for cloning using the procedure of mutually primed synthesis (Oliphant et al., 1986). The random oligonucleotidesused to construct the Z replacement library were constructed from a single-stranded oligonucleotide with the where N represequence 5’-GGGAATTC-N15-CCGAATTCGG-3’, sents an equal probability of all four bases. Following second strand synthesis and restriction with EcoRI, the oligonucleotide was cloned into the EcoRI site of YIp55-his3-GG282(Brandl and Struhl, 1990) (Fig. l ) , to generate his3-GZ derivatives that replace the native Z element with random sequences. Mutations in the his3-GZ-J18,565, and 577 alleles were similarly prepared from oligonucleotides containing an 80% bias toward the wild-type sequence or from oligonucleotides that contained more directed sequence changes. Derivatives of the his3-GZ alleles lacking the upstream REB1-binding site (i.e. his3-GZAR) were constructed by replacing their gall-IO sequences with those from YIp55-his3-GG310 (Fig.1,Brandl and Struhl, 1990) which contains a linker disruption of the REB1-binding site. Derivatives lacking the GCN4-binding site (his3-AGZ and his3-AGG155) were constructed from the related his3-GZ alleles with a oligonucleotide that converted the binding site to the nonfunctional sequence GGACTCT (Hill et al., 1986). Z elements were placed downstream of a GCN4-binding site in his3-ATZ derivatives by cloning the relevant EcoRI oligonucleotide fragments at the junction between upstream and downstream end points from Sc2884 and Sc3111, respectively (Struhl, 1982, see Fig. 4). This places the Z elements at -24of his3, with his3 elements upstream of -80, including the GCN4-binding site directly upstream. The related control his3-AT-Tr (Tr- and T F ) derivatives with TATA elements downstream of GCN4 were constructed by inserting EcoRI-KpnI fragments containing the Tr and the his3 structural gene from YIp55-Sc3640 and Sc3686 (Chen and Struhl,1988) into YIp552884 (Struhl, 1982). Selection of Z Elements-DNA elements that would activate transcription in theabsence of Z where initially selected from his3 alleles integrated into the constitutive GCN4 expressing strain KY322 (relevant genotype, urd-52 his3-A200, Brandl and Struhl, 1990) and then verified as the gene replaced alleles. Briefly, a library containing approximately 10,000independent clones was constructed by cloning random 15-mers into the EcoRI site of YIp55-his3-GG282 (Brandl and Struhl, 1990). These YIp55-his3-GZ molecules were linearized with XbaI and integrated into KY322 using standard protocols (Ausubel et al., 1990). Approximately, 1200 ura+ clones were analyzed for their relative expression as measured by their growth on minimal plates containing 10 mM aminotriazole (AT), a competitive inhibitor of the his3 gene product. The clones having a growth rate above background, approximately lo%, were selected for gene relacement after growth on 5-fluoroorotic acid (Boeke et al., 1984) and retested for his3 expression. The his3 promoter regions from the BamHI to Sac1site of fourteen his3-GZ alleles still expressing above background were recovered by amplifying genomic DNA using the polymerase chain reaction. The recovered fragments were sequenced, cloned back into YIp55-his3-GG282, then gene replaced again into KY322 where their phenotypes were verified. Phenotypic Anulysis-DNAs representing the various his3 alleles were introduced by gene replacement into yeast strain KY322. Growth of the resulting strains was tested in the presence of 10 mM AT. As shown previously, growth in the presence of AT is directly related to the level of his3 mRNA (Chen and Struhl, 1988; Hill et al., 1985). While the absolute growth in AT is dependent upon the background promoter elements and carbon source, the data have been standardized for each set of derivatives. For his3-GG1-related molecules these have been defined as +++ and are his3-GG155 (Brandl and Struhl, 1985) (a 5’ truncated derivative of his3-GG1) and his3-GZ-Jl8 (a his3-GZ derivative with a TATA element upstream of GCN4) for strains grown in glucose and galactose, respectively. The phenotypes were defined as follows: ++++, grows better than thestandard on 10 mM AT; +++, grows as well as the standard on 10 mM AT, ++, reduced growth on 10 mM A T +, poor growth on 10 mM A T +/-, barely detectable growth on 10 mM A T -, no growth on 10 mM AT. These levels of his3 mRNA and the transcriptional initiation sites

Promoter-specific TATA Sequences were confirmed using oligonucleotide probes and S1 mapping techniques described previously (Chenand Struhl, 1988). Purification of Recombinant YeastTBP and DNA BindingAssaysThe coding sequence of Saccharomyces cereuisiae TBP was inserted downstream of the NdeI site of pT7-7tt (Tabor, 1990; generously provided by Drs.S. Tabor and L. Mattheakis). The resulting plasmid was transformed into Escherichiacoli strain BL21 (DE3),and expression of TBP was induced with IPTG as described (Rosenberg et al., 1987). After induction, cells (100mlof culture at ODm = 1.0) were grown at 30 "C for 3 h, harvested by centrifugation (4,000 X g, 10 min), washed once with lysis buffer (20 mM Tris-C1, pH 7.9,at 4 "C, 1mM EDTA, 10% sucrose, 0.5 M NaCl) and repelleted. The cells were resuspended with 4 ml of lysis buffercontaining 0.5 mg/ml lysozyme, incubated at 0 "C for 15 min,and disrupted by sonication. The lysate wasclearedby centrifugation at 25,000 X g for 20 min, and the resulting supernatant was adjusted to a final concentration of 0.3% polymin P, stirred for 15 min at 0 "C, and recentrifuged (10,000 X g, 30 min). The supernatant was adjusted to 55% saturation with ammonium sulfate, stirred at 0 "Cfor30 min, and the proteins pelleted (10,000X g, 30 min). The pellet was resuspended in a volume of buffer T (30 mM Tris-HC1,pH7.9 at 4 "C, 2 mM EDTA, 20% glycerol, 1 mM DTT, 1 mM PMSF) such that the conductivity was equivalent to 0.25 M KCl, and the suspension was applied to a 1.5-ml heparin-agarose column equilibrated with buffer T containing 0.25 M KCl. After washing withthe same buffer,TBP activity, measured by in vitro transcription (Wobbe and Struhl, 1990), was eluted with a linear gradient of0.25-0.6 M KC1 in buffer T. The peak fractions were pooled, dialyzed against buffer T, 0.1 M KCl, and applied to a MonoS fast protein liquid chromatography column(Pharmacia LKB Biotechnology Inc.).Fractions containing TBP activity, obtainedby elution with a linear gradient of0.1-0.6 M KC1 in buffer T, were pooled and dialyzed against buffer T, 0.1 M KCl. This fraction containing TBP at greater than 90%homogeneitywasstored at -80 "C. DNA probes of 71 bp spanning the Z elements were prepared with 32Pend-labeled primers (sequences 5"CTTCGCGAGCTCAAAAAGAGTC-3' and 5'-TTACTGAAAGTTCCGGAATTC-3') that were used in polymerase chain reactions with the YIp55-55-his3-GZ alleles as templates. These probes were further purified by electrophoresis on 8% polyacrylamide gels and eluted as described (Sambrook et a[., 1989).TBP binding was measured bya gel mobilityshift assay carried out as described previously under conditions where TFIIA was not required to observe a TBP complex (Maldonado et al., 1990). Radioactivity present in bound and freeDNAwas quantitated using a Phosphorimager (Molecular Dynamics).

20945

\

/

1.

I

0 6W

sal

I

OUlIll 0 4w

GC310

300

FIG. 1. Structure of his3-GG166,282, and 310. his3-GG155 is a derivative of his3-GG1 (15) which lacks nonessential sequences between -447 and -170 of his3 and 11 bp at the 5' end of gall-10. The gal box contains gall-10 sequences from 649 to 299 (Johnston and Davis, 1984) with four GAL4-bindingsites (boxed) (Bram et al., 1986; Giniger et al., 1985). This region has been fused between -447 (BamHI) and -24 of the his3 promoter(Struhl, 1985) and is flanked by an EcoRI-Sac1 oligonucleotide of 26 base pairs that contains the his3 wild-type GCN4-binding site.The REB1-bindingsite, formerly called Q, overlaps the second most proximal GAL4-binding site and is positioned 127 bpfrom the +1 site of transcriptional initiation (Brandl and Struhl, 1990). The Z element is centered approximately 60 bp from +1 (Brandl and Struhl, 1990). his3-GG282 isa derivative of his3-GG155 which contains gull-IO sequences from 649 to 320. Deletion of these 21 bp results in loss of Z function (Brandl and Struhl, 1990).his3-GG310has a 12-bpinsertionbetweengall-IO sequences 399-391 that disrupts the REB1-binding site and results in a loss of promoter function.

ments directly adjacent to the GCN4-binding site, thisposition is equivalent to that found ina functional derivative of his3-GG155,where 12bparedeleted between Z and the GCN4-binding site (his3-GG348, Brandl and Struhl, 1990). Nine alleles were selected at random for sequencing toverify the random natureof the cloned oligonucleotides. These had a base compositionof 28% G, 27% A, 21% T, and 24% C. The library of his3-GZ alleles was integrated into the chromosome of yeast strain KY322, and the phenotypesof approximately lo3 clones were analyzed in a three-step process. Clones that grew on minimal plates containing 10 mM AT with glucose as RESULTS the carbon source were selected for gene replacement after Random Selection of Functional Replacements for the Z growthon 5-fluoroorotic acid. The phenotypes were confirmed, then the his3 alleles were recovered through use of Element-Wepreviously characterized a gall-l0/his3 chithe polymerase chain reaction,sequenced, and re-introduced meric promoter, his3-GG1, which contains gall-10 regulatory sequences flanked downstream by a GCN4-binding sitewhose by gene replacement into KY322. Fourteen elements that could functionally replace the Z elementandtherelative expression is comparable to the basal his3 promoter (Chen promoter strength of their corresponding his3-GZ alleles, as and Struhl, 1989; Brandl and Struhl, 1990). This promoter 10 AT and a truncated derivative his3-GG155 (Fig. l),unlike gal/ measured by growth on minimal plates containing mM with either glucose or galactose as the carbon source, are his3 hybrid promoters that depend upon a TATA element (Chen and Struhl, 1988), is expressed in glucose media and shown in Table I. Growth in the presence of AT is directly does not require GAL4. Expression of his3-GG155 requires related to the level of his3 mRNA and provides a sensitive the GCN4-binding site,a REB1-binding site, and an element assay for changes in expression (Hill et al., 1986; Chen and termed Z, defined by deletion and disruption analyses to be Struhl, 1988). For the alleles studied in detail, his3-GZ-Jl8, approximately 60 bp from +1 (Chen and Struhl,1989; Brandl 565, and 577, this was confirmed by determining his3 mRNA (Fig. 2). and Struhl, 1990, see Fig. 1).The expression of his3-GG155 levels and transcriptional initiation sites Sequence and phenotypic comparisons suggested that the thus serves as amodel system for a regulated TATA-less elements could be divided into three functionalgroups. Three Z element, we choseto promoter. Tofurtherdefinethe genetically select from a random library those elements that of the elements(518,B9, and C28, these will be referred to as been previously defined can restore expression. This random selection approach was Group 1)contain sequences that have used t o facilitate the identification of the range of elements as functional TATA elements as they allow transcriptional that would function for Z as well as to potentially optimize activation in conjunction with an upstream activator (Chen and Struhl, 1988; Harbury and Struhl, 1990). This was conthese elements without bias toward the native sequences. A library of approximately lo4 his3-GZ alleleswas con- firmed by the ability of these elements to supportgrowth in GAL4structed by cloning oligonucleotides containing 15 randomized galactose mediainconjunctionwiththeupstream binding sites at a rate only marginally slower than a gallhis3 bases flanked by EcoRI restriction sites into the EcoRI site of YIp55-his3-GG282, a truncated derivative of his3-GG155 chimera containinga wild-type TATA element inplace of the GCN4-binding site (his3-Gl7, Chen and Struhl, 1988). Seven that lacks 20 bp upstream of the GCN4-binding site and consequently Z function (Fig. 1). While this places the ele- elements (Group 2) contained the sequence G2_3TPuPu. (An

Promoter-specific TATA Sequences

20946

TABLE I Sequence and phenotype of the randomly selectedZ elements His3-GZ alleles in yeast strain KY322 that facilitated growth on minimal plates containing 10 mM AT and 2% glucose were sequenced. Phenotype refers to growth rate on plates containing 10 mM AT and either 2% glucose or galactose as the carbon source. Growth rate is standardized as +++ for his3-GG155 and his3-GZ-Jl8 for cells grown on glucose and galactose, respectively. For a further definition of scoring see “Materials and Methods.” All Z oligonucleotides are flanked on up and downstream sides by EcoRI sites (lowercase). Bases in boldface type were not randomized due to restrictions of the cloning procedure. Elements are classified into Group 1.2, or 3 asdefined in the text on the basis of sequence similarity and phenotype. A5 is a similar EcoRI flanked oligonucleotide containing the native Z element. The TATA sequences of Group 1 and TATA-like sequences of the Group 2 elements are underlined with a solid line. The GGTPuPu sequences of the Grow2 elements are underlined with a broken line. Phenotype

his3-CZ Sequence allele

B427 B424

B425 B426 B453

DNA

Group

J18 C28 B9

B471

1 1 1

565 G41 G30 577 H6 H50 K14 D57

B473 B454 B455 B472 B470 B469 B535 B456

B21 C25 D87 B655

g a a t t c GG AGTAGCATATAAATC g a a t t c CCGGCTATAAACGTCC CC ACATATATTTAGTAT CC

GG GGGGT-WTAAAGCA GG GAWiTGGTATAGTA GG CGGAGGNTM-TTAA IGAGAGATTGTACAA

@ - JGAGAGAGATTAATA GG T T G A G G G A G G E GG-~T~.GTGGGCTTTAAGA

3 3 3

A5

Galactose Glucose

+++

+ ++ ++++ ++ ++ +++ ++ + ++

GG GG&GCAAGGCTA

+I-

GG AGAGGITXGAAA

+I-

GG ATTGGGCCATGCAAG GG GAAAGGACAAGAG TGAAAGTTCCAAAGAGAAG

+

+ +++

+++ ++ +++

-

preference rather than selection as seven of nine elements sequenced a t random had this orientation. Importantly, however, the G2-3TPuPu similarity could be extended in the 565 +1 and G41 elements to G6TAPu except for a single mismatch in the upstream G stretch. 577 and H6 were also strikingly h is3 similarwith a completematch over theirfirst 7bp not including the 5”GG dinucleotide. Furthermore, unlike Group +12 1, none of the Group 2 elements contained TATA elements that would allow GAL4-inducedactivation. Three of the weaker elements (B21, C25, D87; Group3)containedsequences that resembled the native Z element. The closest match occurs in D87 where 10 bases (sequence GAAAGGACAAGAG) could be aligned with the Z element in gall-10 sequences from 328 to 310 (Johnston and Davis, 1984) provided that a central gap (not underlined) wasintroduced. Although not closely related, B21 and C25 also shared either GAAAG or CAAG sequences and did not promote growth in galactose medium. 565 J l S 2 8 2 155 577 Z Elements Require GCN4- and REBl -binding Sites for Full FIG. 2. RNA analysis. RNA was prepared from the GCN4 exActivity in the his3-GZ Promoters-Z elements 565, 577, and pressing strain KY322 containing the indicated his3-GG or GZ alleles. Cells were grown in minimal media with 2% glucose. For each 518 were selected fordetailed analysis. These were the strongsample, 25 pg of total RNA was hybridized to his3 and dedl 5’ end- est elements and appeared to represent two distinct groups: labeled oligonucleotide probes, digested with S1 nuclease and after 518, the TATA containing Group 1, and 565 and 577, the separation on a denaturing acrylamide gel, subjected to autoradiog- GGTPuPu containing Group 2. T o ensure that these elements raphy. Lanes: I , his3-GZ-J65; 2, his3-GZ-Jl8; 3, his3-GG282; 4, his3- were activatingtranscriptioninthesamemannerasthe GG155; 5, his3-GZ-577. +I and +I2 start sites for his3 mRNA are native element, derivatives were constructed to determine if indicated. their activity required GCN4 and REBl binding sites. his3GZAR derivatives were constructed from his3-GG310 (Fig. 1) eighth element, H50, with the sequence GGTTGA was in- to generate alleles that contain a linker disruption of the cluded in this group.) All of these elements were oriented in REB1-binding site. Loss of the REB1-binding site reduces the same direction as shown by the GG dinucleotide flanking transcription approximately &fold and prevents growth on the upstream EcoRI site which for two of the elements con- plates containing 10 mM AT in the context of the native Z tributes in part to the consensus. These two bases were not element (Brand1 and Struhl, 1990; his3-GG310, see Fig. 3). randomized due to requirementsof the mutually primed syn- The equivalentZ element containingalleles, his3-GZAR-518, thesis cloning (see “Materials and Methods”). I t should be 565, and 577, also did not permitgrowth on 10mM AT plates noted that this preference may partially result froma cloning confirming their requirement for the REB1-binding site. To

-

-

Promoter-specific TATA Sequences

20947

518-21, 518-71), as shown by the reduction in GAL4-activated transcription, or inverted its orientation (J18-4r), also reduced GCN4/REBl-activated transcription inglucose. This confirmed that the TATA element in 518 was necessary for the expression of his3-GZ-518. Two mutations outside the TATA element (518-62 and 518-83) did not decrease either GAL4- or GCNli/REBl-activatedtranscription. However, mutation of A to T at the third position of 518-84 reduced the level of expression in glucose without alteringthe expression in galactose. This was the first indication that while transcriptional activation by upstream or downstream positioned activator proteins can mediated be by a TATA element, the flanking sequence requirements of the TATA element is dependent upon the relative position of the activator. This likely explains why, in ourprevious study (Brandland Struhl, 1990), the directed placement of a TATA elementin the position of Z did not activate transcription. An alternative argument that functional differences are solely the result of TATA strength is less likely because B9 and 518 induce GAL4 expression to the same extent yet differ with regard to Z activity. Group 2 Elements Contain Weak TATA Sequences-Although none of the Group 2 Z elements activated transcription of their respective his3-GZ derivatives in conjunction with GAL4 in galactose medium, the fact that TATA elements would function for Z prompted us to test if they had TATA activity in conjunction with an upstream GCN4 site. In addition, many of the Group 2 elements including 565 (AAATAAA), G30 (TAATTAA), and K14 (TTTAAG) contained sequences that resembled a TATAelement. A failure to detect TATA activity from these elements may have resulted from the inability of GAL4 to respond to certain TATA sequences that can be productive with GCN4 (Harbury and Struhl, 1990).Z elements were cloned downstream of a GCN4-binding site in a deletion derivative of the his3 promoter that lacks sequences between -24 and -80 and thus also both native downstream TATA elements, T r and Tc (Struhl, 1986; Fig. 4A). In these his3-ATZ derivatives, the Z elements can be assayed for conventional TATA activityin combination with GCN4 and compared to TATA sequences that have been 2-5‘17 ACCISS analyzed previously (Harbury and Struhl, 1990; his3-AT deCZz\H..l65 \G%-J65 rivatives differ from his3-A93 only in the placement of the FIG. 3. Growth of strains containing his3-GZ alleles. Yeast TATA element a t -24 rather than -32). As expected, the strain KY322 cells containing the indicated alleles were streaked onto TATA element containing Z elements 518 and C28 facilitated minimal plates containing 10 mM AT and 2% glucose and grown at the activated expression of this promoter at levels comparable 30 “C for 4 days. his3-GZAR and his3-AGZ derivatives lack REB1- to the his3 wild-type element (his3-AT-TP) (Fig. 423). Two and GCN4-binding sites, respectively. The comparable derivatives for his3-GG155are his3-GG310 and his3-AGG155. his3-GG282which Group 2 elements, 565 and G30, did in fact activate transcriplacks the Z element was included as a a negative control and yeast tion in the his3-ATZ promoter configuration at a level that enabled growth on 10 mM AT. Although these did so at a strain KY114 with a wild-type his3 gene as a positive control.

examine the requirements of the GCN4-binding site, point mutations were made to convert ittothe nonfunctional sequence GGACTCT (Hill et al., 1985) in thehis3-AGZ derivatives. The three alleles his3-AGZ-518, 565, and 577 all resulted in a reduction in growth rate similar to that observed when the GCN4-binding site was disrupted in the contextof the native element (his3-AGG155). While only his3-AGZ-577 was as severely affected asthe native Z element(his3AGG155), it should be noted that theGCN4-binding site will partially repress TATA-dependent expression from an upstream activator sequence (Brandl and Struhl, 1990). Removal of the GCN4-binding site consequently results in an increase in activation by REBl which is a weak activator in conjunction with a TATA element (Chasmanet al., 1990; Finley and West, 1989), due to relief of repression. As shown below 565, like 518, does possess TATA activity. This complexity aside, the requirements for REBl andGCN4 are very similar if not identical, for transcription of the his3-GZ alleles as for his3GG1. A TATA Element Can Substitute for 2-The Group 1 elements, 518, C28, and B9, facilitated GAL4-mediated transcription of their respective his3-GZ promoters in galactose media. Not surprisingly, all three contain sequences that have previously been defined as TATA elements; these are TATAAA in the case of C28 and 518 and TATTTA in the case of B9. To determine if the TATA elementwas responsible for the Z activity, mutations of hk3-GZ-518 were analyzed. All of these derivatives contain a G to T transversion at the 5‘ end of the element which has no effect on function. The phenotypes of these derivatives are shown in Table 11. Mutations which disrupted the TATA element (518-5, 518-12,

TABLE I1 Sequence and phenotypes of his3-GZ-JI8 mutations

Yeast strain KY322 cells containing his3-GZ alleles with the Z element sequences shown were grown on minimal plates containing 10 mM AT and either 2% glucose or galactose as the carbon source. Growth rates are defined as in Table I and described under “Materials and Methods.” Mutated bases relative to his3-GZ-J18-4 are underlined. As is the case for all Z elements, the sequences are flanked by EcoRI (Rl) restriction sites. his3-GZ allele

DNA

518 518-4 518-5 518-12 518-21 518-71 518-62 518-83 518-84 J18-4r

B471 B607 B608 B610 B502 B543 B532 B527 B534 B605

Sequence

Rl

Phenotype Galactose

aAGTAGCATATAAATC

R1

TG AGTAGCATATAAATC TGAGTAGCAFAAATC TG AGTAGCATCTAAATC TGAGTAGCeTAAATC TG AGTAGCATACAAATC TG A-TAGCATATAAATC TGAGFGCATATAAATC TGIGTAGCATATAAATC GATITATATGCTACT CA

Glucose

+++ +++

++ ++ + +++ +++ ++ + +/-

+++ +++ ++/-

+++

++++ +++ +

Promoter-specific TATA Sequences

20948 A -90

-110

I

dAdT

-70

-50

-30

-10

GCN4

B his3 -AT

AAATAAA, of 565 was important for its function, a series of mutations of his3-GZ-565 were examined for activity (Table 111).Mutations of the 565 element, 565-1,4, and 16, decreased Z activity and indicated that the centralbases making up the sequence AAATAAA were necessary for Z function. As all of these mutations also decreased activity of the corresponding his3-ATZ promoters and TBP binding of the elements (see Fig. 5 ) ,this sequence does possess TATA activity and is likely functioning as an interaction site for TBP. Again, however, it should be emphasized that 565s level of Z activity does not correlate directly with its TATA element activity suggesting a significant role for sequences outside the TATA element core. ~ n ~ r Two other mutations are of interest. Atransition of G to A in the G stretch of 565-2 resulted in reduced Z and TATA activity. Presumably this was the result of reduced binding of TBP (see below) and identifies a position outside of the core TATA that affects TBP binding. The T to C change of 56545 reduced TATA activity in his3-ATZ without affecting Z activity. This mutation also resulted in a 60% reduction of TBP binding (see below) and suggests that while TBP binding maybe required for Z activity, the his3-GG(Z) promoter configuration can tolerate a lower affinity binding of TBP than thehis3-ATZ configuration. Sequences Outside the Core TATA Are Critical to Z Actiuity-The Group 2 element 577, displayed weaker TATA activity and TBPbinding activity than 565 (see Figs. 5 and 6 ) , yet had almost equivalent Z activity. To confirm that this element contained TATA-like sequences and identify additional sequences that contributed to Z activity, a series of mutations of his3-GZ-577 and his3-ATZ-577 were tested for activity (Table 111).The potential TATA sequence TACAAG is present at the 3' end of the element. Two elements, 57713 (compare with 577-37) and 577-19, with mutations within this sequence reduced expression, while a C to T transition in 577-2 resulted in an increase in expression. This latter change makes the sequence a better TATA element(TATAAG) as witnessed by its activity with GCN4 upstream in the his3-ATZ configuration and TBPbinding. The role of the TATA-like sequence became more evident as a further improvement in the TATA (577-26) resulted in enhanced activity when growth rates of strains his3-GZ-J77,577-2, and 57726 were compared on plates containing40 mM AT. The results with 577 are even more striking than with 565 as they show that a weak TATA sequence that functions only poorly as a downstream TATA element can function when positioned upstream of a GCN4-binding site. In the background of a weaker TATA sequence, the 577 element allowed the identification of additional sequences that contribute to Z activity. Mutations within the G2TPuPu sequence conserved in the Group 2 Z elements (577-29 and 32) dramatically decreased growth on 10 mM AT, indicating thatthis sequence conservation was not coincidental. It should also be noted that thestrongest Group 1 element, 518, contains at its 5' end the variation of this sequence, GAGTAG. In this case, mutation to GTGTAG in 518-84 resulted in reduced Z element activity. In addition, mutations within sequences conserved between 577 and H6 (577-9 and 577-37) resulted in decreased growth on 10 mM AT. Allof these mutations lie outside the core TATA-like sequence yet decrease Z activity. As the 577 element had little TATA function, to examine whether mutations in the 5' portions of the element were promoter specific, elements were synthesized that contained the upstreammutations of 577-32 in the context of the improved TATA elements of 577-2 and 577-26. These were examined in both his3-GZ and his3-ATZ promoter arrangements. The upstream mutations reduced expression even in

z his3

DNA

allele

Potential TATA sequence

Phenotype

P ~ ~ ' ~ ~

his3-AT his3

++ +

-GZ

Z 565

B574

TAAATAAA

G30 G41

B561 B596

TAATTAA TATAGTA

577 C28

B590 B570

TTGTACAA CTATAAAC

+++

t

518

B592

ATATAAAT

++++

+++

Tr wt 217

B540

TATAAA TATAAG

++++

B542

+++t

++ ++ +++

na na FIG. 4. Structures and phenotypesof his3-AT Z and AT-Tr derivatives. Panel A , diagram of the his3 promoter with the position of the poly(dA-dT) sequence, GCN4-binding site, and Tr and Tc TATA elements indicated with respect to the +I mRNA initiation site (Struhl, 1985, 1986). Shown below are the junctional sequences of the ATZ and AT-Tr derivatives with the GCN4-binding site underlined and theposition for insertion of Tr or Z elements indicated by bold type. Z elements were inserted between an 8-bp EcoRI linker. Tr elements were inserted at thejunction of EcoRI and Sac1 linkers. Panel B, resulting phenotypes of the his3-ATZ (or AT-Tr) alleles containing the inserted sequences shown are compared with their equivalent his3-GZ alleles. Strains weregrown on minimal plates containing 2% glucose and 10 mM AT with growth rates being compared to his3-GG155 which was scored as +++ (for details of scoring see "Materials and Methods"; n u , not applicable).

+

comparatively low level, similar to the TATAAG sequence of h i ~ 3 - A T - T r " (Chen ~ and Struhl, 1988), this finding suggests that the Group 2 elements do possess conventional TATA element activity and that the distinction between Group 1 and Group 2 elements may solely reflect the level of this activity. Mutational analysis of the 565 (shown below) identified the TATA-like sequence AAATAAA as essential for activity in the his3-GZ and his3-ATZ promoter configurations. In addition, weak TATA activity could be seen with the other elements when growth rates were compared on minimal plates versus an allele with no element present (not shown). The Group 2 Z elements thus possess some, albeit weak, conventional TATA activity. Since the strongest Z elements include 565 and 577, these results have a second important implication; that is, they clearly show that there is not a direct correlation between conventional TATA element activity (transcription of his3A T Z ) and Z element activity (transcription of his3-GZ). TATA element strength contributes to, but is not the only determinant of Z activity. As shown with the 518-84 element and below with mutations of the 577 element, sequences outside the TATA element core play acritical role in Z element function. The range of TATA sequences that will function in conjunction with an activator, in this case GCN4, is therefore dependentupon the context inwhich the activator is found; this may include its placement with regard to other elements as well as what other elements are present. The Sequence AAATAAA Is Required for the 2 Element Actiuity of 2 J65"To confirm that theTATA-like sequence,

Promoter-specific TATA Sequences

20949

TABLE 111 Sequence and phenotypeof his3-GZ and his3-ATZ-J65 and 577 mutations Yeast strain KY322 cells containing his3-GZ or his3-ATZ alleles with the Z element sequences shown were grown on minimal plates containing 2% glucose and either 10,20, or 40 mM AT. Growth rates are defined as in Table I and described under “Materials and Methods.” Mutations relative to the starting565 or 577 elements are underlined. The DNA number in parentheses refers to the hisd-ATZ allele. Phenotwe Element

Phenotwe his3-GZ

DNA

Sequence

565 J65r

B473 (574) B556

GGGGGGTAAATAAAGCA TGCTTTATTTACCCCCC

565-2 565-16 565-45 565-4 565-1

B555 (711) B557 (710) B575 (729) B584 (712) B589

GGG5GGTAAATAAAGCA GGGGGGTAAATSAAGCA GGGGGGSAAATAAAGCA

577 J77r

B472 (590) B566

GGTGAGAGATTGTACAA TTGTACAATCTCTCACC

577-19 577-9 577-13 577-37 577-29

B567 B558 B560 B595 B727

GGTGAGAGATTSASAA GGTGAGAG-TITTACAA GGTGCjGAGATTGTw GGTGFAGATTGTACAA GGSGAGAGATTGTACAA

+ ++ + ++ +

577-32 577-2 577-46 577-26 577-40

B599 B553 (730) B751 (741) B726 (743) B740 (742)

GGTGSGIGATTGTACAA GGTGAGAGATTGTAIAA GGTGSGIGATTGTATAA GGTGAGAGATTATATAA GGTGSGTGATTfiTATAA

++++ +/++++

10 A T

Eklllent: J18

-5

-4

hisSGZ

7: +++ +++ ++ Gal: +++

his3 \Tz:++++

+++ + NO

NO

-

+I-

+I-

-+

++

+-

++

-84

J65

++

++++ +++

+++ NO

-2

-

-

++

+

-16

-45

++ ++++

-

.

-

+

40 A T

++++ +++ ++ ++++ ++ +++

GGGGGGTAASTAAAGCA GGGGGGTAA-TTAAAGCA

-12 -71 -21

his3-AT2

+

10 A T

20 A T

++

+

40 A T

-

-

-

++++ + ++++

++++

-

-

++ ++++ ++ +++ ++++ ++++ ++++

+

-4

-1

J77

-

++

+++

-

-

NO

-

-19

-9

-13

-37

-29

++

+

+

++

+

NO

NO

NO

NO

ND NO NO

-

-

NC NO

-32

-

NO

-2

++

-46

++++ + +

-

-26

-40

+ ++++ ++ ++++

++++ ++ ++++ ++++

A5

+++

-

NO

FIG.5. TBP binding of mutations of the 518, 566, and 577 Z elements. Mobility shift assays were performed as shown in Fig. 6 and described under “Materials and Methods.” The affinity of the indicated Z elements (sequences are found in TablesI1 and 111) is expressed as percent of probe shifted into the TBP.DNA complex and were quantitated using a Phosphorimager (Molecular Dynamics). The growth rates of strains containing his3-GZ and his3-ATZ alleles with these Z elements, on minimal plates containing2% glucose or galactose and 10 mM A T (see Tables I1 and I11 and Fig. 4) are summarized below the graph. The results with an oligonucleotide containing the native Z element sequence (A5, Table I) is also shown. ND, not determined.

the context of the improved TATA sequences of 577-2 and 577-26 (compare with 577-46 and 577-40, respectively). In fact, the upstream sequences were more critical for expression than the TATA sequences; that is, a strong TATA sequence could not compensate for mutationswithin the upstream sequences (compare 577-40 with 577 and 577-2). When the Robe* same elements were placed into the his3-ATZ promoter arrangement, it was apparent that the same upstreammutations mID: reduced transcription (compare 577-26 and 577-40 on 40 mM Oaent Jl8 JTT &S Mt c90 H5O K14 D87 821 C25 D57 AT); however, in this case the upstream sequences were not FIG. 6. TBP mobility shift assay of therandomly selected Z elements. Radiolabeled DNA fragments from the indicated his3-GZ as significant as the TATA sequence (again compare 577-40 alleles were incubated in the absence (-) or presence (+) of 50 ng of with 577 and 577-2). Thus, while the upstream mutations are recombinant yeast TBP. Free probe (probe) and TBP-probe com- not totally promoterspecific there must be a difference in the plexes (ZID. DNA)were resolvedon polyacrylamide gels as described transcription of the his3-GZ and his3-ATZ promoters such under “Materials and Methods” and visualized by autoradiography. that the effect of these mutations is minimized when the elements are used as downstream TATAelements.

” - t - t - t - + - + - + - + - + - + - + - +

” ” ” ” “ _

20950

Promoter-specific TATA Sequences

577-2 M Z Element ActivityRequires a Lower T B P Binding Affinity Than Conventional TATA Activity-The finding that certain 577-46 C 577-2 C ., TATA elements could function as Z elements and that the Group 2 elements had weak but detectable TATA activity suggested that the Group 2 elements were binding sites for TBP. To determine the level of this binding in vitro, bacterially expressed yeast TBP was used in mobility shift assays with Z element containing DNA fragments in the gel conditions of Maldonado et al. (1990) which eliminate the requirement for TFIIA (Figs. 5 and 6). Binding of TBP, expressed as a percentage of the probe shifted, varies from approximately 18% for 518 to 1%for 577. This level of binding for 518 was approximately 50% that seen for the strong adenovirus major late TATA element and 2-fold that seen for the 511 M his3 wild-type TATA element (G17, Chen and Struhl, 1988) FIG.7. Growth of strains containing hi83-GZ alleles in the (not shown).565binds at an intermediate level, approxi- presence of an altered binding specificity mutant of TBP. mately 5%, while less than 1%of the native Z element (A5) Yeast strain KY322 cells containing the indicated his3-GZ alleles was bound. A clear distinction was observed when compari- were streaked onto minimal plates containing 10 mM AT and 2% sons of TBP binding vers'sus Z and TATA activity were made glucose and grown a t 30 "C. M indicates those cells containing a among the Z element groups. TATA activity, the ability of centromeric plasmid expressing the altered specificity mutant of the element to activate transcription inconjunction with TBP, M3 (recognition sequence TGTAAA;Strubin and Struhl, 1992) while C indicates those cells containing a control plasmid. GAL4 in his3-GZ or with GCN4 in his3-ATZ, correlated directly with TBP binding. In contrast, the Z activity of the three groups did not correlate directly with the i n vitro binding (data not shown). Again this suggests that the his3-GZ promoter configuration is less sensitive to thebinding affinity of of TBP; 565 and 577 bind TBP relatively weakly yet are strong Z elements. Z element activity can seemingly tolerate TBP. a lower affinity TBP binding, as measured in vitro, than DISCUSSION conventional TATA activity. TATA elements will facilitate transcriptional activation on The TBP binding of 518, 565, 577 and their derivatives is shown in Fig. 5. Within any one set of derivatives, in the either side of an activator protein. Classically, TATA elecontext of the same flanking sequences, there was a general ments have been considered as downstream promoter elecorrelation between T B P binding and both Z and TATA ments, positioned 3' of an activator protein-binding site and activity; for example, compare 518 with 518-12 and 518-21, proximal to thestart site of transcription. With the exception 565 with 565-4 and 565-31,577 with 577-2 and 577-26. The of a group of promoters that have been considered as TATAlast example correlates increased binding with increased ac- less, this is the principal structure for regulated RNA polymtivity. This agrees with TBP being the functional agent of Z erase I1 promoters in S. cerevisiae. By selecting random seactivity. The correlation was,however, tighter for TATA quences that functionally replace one of the required upstream activity than Z activity. As seen for some of the 565 mutations, elements of the "TATA-less" his3-GG1 promoter (Chen and similar TBP binding capabilities resulted in different Z ac- Struhl, 1989; Brand1 and Struhl, 1990), the Z element, we have shown that TATA elements can activate transcription, tivities (for example, compare 565-4 and 565-16). The effect on TBP binding of flanking sequence mutations when positioned upstream of a GCN4-binding site. Of 14 that result in decreased Z activity was of particular interest sequences selected to function upstreamof GCN4, three were based upon, 1) their ability to and was most easily seen by comparing 518-84 with 518 and certainlyTATAelements activate transcription in conjunction with GAL4 or an up577-40 with 577-26. For both of these pairs, the mutated element bound TBP aswell or better than their nonmutated stream GCN4, 2) their sequence characteristics, and 3) their counterpart. This demonstrates that the sequences flanking strong binding of TBP in vitro. This finding indicates that the core TATA element must contribute a component to Z there is no inherent restraint imposed by the mechanism of transcriptional activation that prohibits the positioning of an activity distinct from TBP binding. Z Element Activity Is Enhanced by a n Altered Binding activator between the TBP-binding site and the transcripSpecificity Mutant of TBP-The availability of an altered tional start site. Differences in flanking sequence effects may specificity mutant of T B P allowed us to test directly if Z suggest, however, that intricacies of the mechanism may differ element function requires TBP. Amino acid substitutions in depending on the position of the TATA element. In addition, TBP of Ile-194 to Phe-194 and Leu-205 to Val-205 enable it unlike the more typical inverse arrangement, at least in the case of the liis3-GZ and its parent his3-GG promoter, this to recognize the sequence TGTAAA (StrubinandStruhl, 1992). 577-13 contains precisely this sequence at its 3' end promoter arrangement requires an additional upstream element to activatetranscription. For his3-GG(Z) promoters while 577-2 and 577-46 containthe variationTGTATA. These elementsin the context of their his3-GZ alleles as well there is an absolute requirement for a REB1-binding site. Low Affinity TBP-binding Sites Will Activate Transcription as 577 were assayed for activity, as measured by growth on 10 mM AT plates (Fig. 7), inthe presence and absence of the M3 When Positioned Upstream of GCN4-The Group 2 Z elevariant of TBP (kindly provided by Drs. M. Strubin and K. ments were classified as having the sequence GGTPuPu at Struhl). Transcriptionof his3-GZ-577-13 is clearly enhanced their 5' ends and not facilitating GAL4-induced expression. in the presence of the altered specificity TBP. This verifies These elements were weak TATA elements in conjunction that theelement is acting as aninteraction site for TBP. This with GCN4 in the his3-ATZ promoter arrangement and bound activation shows allele specificity as transcription of his3- T B P significantly poorer than Group 1 elements. Despite GZ-577-2 is only slightly enhanced by the mutant while 577 this, with the correct flanking sequences (discussed below), and 577-46 are unaffected. Interestingly, the enhancementof these were the strongest Z elements. If the Group 2 elements transcription by the TBP altered specificity mutant was are functioning via TBP, thehis3-GZ promoter arrangement greater when the corresponding his3-ATZ alleles were tested must tolerate the use of lower affinity TBP-binding sites.

Promoter-specific TATA Sequences Alternatively, the Group 2 elements may bind a factor distinct from TBP thatcan activatetranscription in conjunction with GCN4 in the his3-GZ promoterarrangement. While it is difficult to completely eliminate the latter,we favor the model that theGroup 2 Z elements are acting as interaction sites for TBP based upon the following criteria. First, mutations within the TATA-like sequences of 565 and 577 which reduce TBP binding and TATA activity, also reduce Z activity. Second, mutations that improve the TATA-like character of 577 as witnessed by improved TBP binding in vitroand TATA activity in uivo, show a corresponding increase in Z activity. Third, theGroup 1elements reveal that TBP has potential the to function in an upstream position. Fourth, the activity of selected Z elements, 577-13 in particular, are enhancedin the presence of an altered specificity mutant of TBP that binds preferentially to its sequence. While TATA elements could functionally replace Z, the fact that therewas not a precise correspondence between the Z and TATA element activity of the Group 1and 2 elements, in addition to the lower in vitro binding of TBP to Group 2 elements suggests that TBP binding affinity is not the only determinant of Z function. It is unlikely that this lack of correspondence is due to spacing differences among the elements since the position of the TATA-like sequences of the Group 2 element are variable. The his3-GZ promoter seemingly can tolerate a much weaker affinity binding site for TBP than theconventional promoter arrangement. What is unique about the his3-GG(Z) promoter arrangement that may accept the use of lower affinity TBP-binding sites? The REB1-binding site,an essential component of this promoter, may bethe critical factor. From a variety of in vitro studies, TFIID is the first basal factor introduced in the formation of the initiation complex (Buratowski et al., 1989: Fire et al., 1984; Maldonado et al., 1990; Reinberg et al., 1987; Van Dyke et al., 1988). Nucleosomes will compete for this binding and a requirement for nucleosome rearrangement that accompanies activation has been suggested from reconstitution studies in vitro (Laybourn and Kadonaga, 1991; Workman and Roeder, 1987; Workman et al., 1991) and from genetic experimentswithalteredhistone levels or forms (Clark-Adams et al., 1988; Han et al., 1988; Han and Grunstein, 1988; Durrin et al., 1991). Any factor such as REBl (Fedor et al., 1988) that could exclude nucleosomes from the TBP-binding siteshould facilitatethe in vivobinding of TBP and eliminate any requirementfor nucleosome rearrangement prior to or simultaneous with TBP binding. In such a promoter arrangement TBP binding would not be in direct competition with nucleosome binding. Indeed, the spacing between the REB1-binding site and the GCN4 and Z elements is critical for the function of this promoter (Brand1and Struhl, 1990). Furthermore, as shown with the series of 577 mutants, increased binding of TBP, in the context of the correct flanking sequences, would still increase expression. REBls ability to promote the activation potential of downstream factors is not limited to TBP. A synergistic enhancement of transcription by appropriately positioned REBlandthe DEDl activator element has also been observed (Chasman et al., 1990). Alternatively, REBl could interact directly with TBP or a TFIID complex to stabilize binding (for examples, see Lieberman and Berk, 1991 and references therein). The combination of certain promoters using TATA elements situated at alternate positions relative to theactivatorbinding site with the ability to use low affinitysites has implications with respect to TATA-less promoters. Obviously, in the context of these parameters it would be difficult to localize the TBP-binding site either by sequence analysis or in vitro approaches that rely upon high affinity binding. Many so-called TATA-less promoters,therefore, probably have

20951

TBP-binding sites that are difficult to localize. This would agree with in vitrotranscription studies (Carcamo et al., 1989, 1990, 1991; Pugh and Tjian, 1990, 1991; Smale et al., 1990) which indicate a requirement for TFIID even in the absence of a TATA element. Similarly, Carcamo et al. (1990) have localized a cryptic TBP-binding site in the TATA-less adenovirus IVa2 promoter, downstream of the transcriptional start site. If, in fact, activator proteins function to stabilize TFIID binding (Sawadogo and Roeder, 1985; Stringer et al., 1990) then promoters with low affinity TBP-binding sites would be the most affected targets (see Lieberman and Berk, 1991; Reach et al., 1991, for examples). Sequences Flanking the TATA Element Affect T B P Function-The Group 2 Z elements were originally classified on the basis of containing a GGTPuPu at the 5’ end of the element. Subsequently, TATA-like sequences were localized within these elements; but the conservation of 5‘ sequences suggested that they play E role in Z element function. Mutation of these sequences in the 577 element confirmed this role. While it could be argued that themutations that disrupt 577 function act by directly or indirectly creating a repressor rather than disrupting a positive function, the fact that upstream sequences were conserved between H6 and 577 make this unlikely. In addition, several mutations had a detrimental effect. We have found that these flanking sequences function without reducing TBP binding in vitro.It is also unlikely that these sequences function solely by enhancing TBP binding in vivo since improvement in the TATA sequences, resulting in a higher level of TBP binding in vitro, will not compensate for their function. Consequently, these flanking sequences identify a process distinct from TBP binding that contributes to the overall transcriptional rate. The effect of the flanking sequences is not promoter specific; that is, mutations within these sequences decrease Z activity in his3-GZ (GCN4 downstream) and TATA activity in his3-ATZ (GCN4 upstream). However, the significance of the flanking sequences relative to the TATA sequences differs in the two promoter types. Flanking sequences are as critical a parameter as the TATA in his3-GZ while the TATA element takes precedence in his3ATZ. This implies that the relative contribution of the ratedetermining steps differ in these promoters (see Colgan and Manley, 1992). Associated factors (TAFs) have been identified which contribute to TBPfunction in higher eucaryotes (Dynlacht et al., 1991; Meisterernst et al., 1991; Pugh and Tjian, 1991; Safer et al., 1991; Tanese et al., 1991; Timmers and Sharp, 1991). Although less tightly associated, similar factors are probably found in yeast (Berger et al., 1990; Kelleher et al., 1990; Flanagan et al., 1991; Koleske et al., 1992). The 5”flanking sequences may act as the recruitment site for such a factor, perhaps in association with TBP. By functioning in a sequence-specific manner, the role of the factor would belimited to a subset of promoters thus providing an additional target for regulation (for potentially related examples, see Lieberman and Berk, 1991; McCormick et al., 1991; Meisterernst and Roeder, 1991; Safer et al., 1991;Fong and Emerson, 1992). Alternatively, the flanking sequences may directly or indirectly enhance the activity of TBP subsequent to binding. For example, any perturbation in DNA structure induced by TBP (Buratowski et al., 1991; Horikoshi et al., 1992) would, at least in part,be influenced by flanking DNA sequences. In our analysis we have grouped Z elements on the basis of sequence and phenotype in galactose media. Groups 1 and 2, however, probably represent a continuum of elements containing various strengths of TATA elements and flanking sequences that in combination generate the observed activity. In thisregard, the 565 TATA-like sequence AAATAAA while

Promoter-specific TATA Sequences

20952

not allowing GAL4 induction in his3-GZ, will allow partial induction in his3-ATZ in vivoand in vitro(Wobbe and Struhl, 1990) and in his3-AGZ invivo (not shown) where GCN4 repression is eliminated. In addition, as noted above, the 5'flanking sequences of 518, the strongest Group 1element, are most closely related to those of Group 2. Perhaps due tolimitations of the random selection approach and the number of sequences that could be handled, we were unable to fully define the sequence requirements of the native Z element. Analysis of additional directed mutants' has shown that theelement lies within sequences from 315 to 329 ofgall-10 (Johnston andDavis, 1984).Our present results suggest that this element may directly or indirectly act as a low affinity recruitment site for TBP, similar to many of the Group 2 elements. Interestingly, Finley et al. (1990), found that thisregion also contains agalactose-derepressed operator sequence. How the protein that mediates this effect relates to Z function will require further definition of its binding site and itsisolation. Acknowledgments-We thank Drs. M. Strubin and K. Struhl for kindly providing the TBP altered specificity mutant, Drs. S. Tabor and L. Mattheakis for pT7-7tt, and Michele Markus for some of the TBP preparations used in this study. We are also grateful to George Chaconas, Michael Surette, BriLavoie, and David Haniford for useful comments on this manuscript, Dr. Arnold Oliphant for computer analysis of the random sequences, and Dr. Kevin Struhl for enabling portions of this work to be done in his laboratory. REFERENCES Ausubel, F. A., Brent, R., Kingston, R. E., Moore, D. D., Seidman, J. G., Smith, J. A,, and Struhl, K. (eds) (1990)Current Protocols tn Molecular Bwlogy, Greene Publishing and Wilely Interscience, New York Berger, S. L., Cress, W. D., Cress, A,, Triezenberg, S., and Guarente, L. (1990) Cell 61,1199-1208 Boeke, J. D., Lacroute, F., and Fink, G. R. (1984)Mol. Gen. Genet. 197, 345346 Bram, R. J., Lue, N. F., and Kornberg, R. D. (1986)EMBO J. 5,603-608 Brandl, C. J., and Struhl, K. (1990)Mol. Cell. Biol. 10,4256-4265 Buratowski, S., Hahn, S., Sharp, P. A., and Guarente, L. (1988)Nature 334, 37-42 Buratowski, S., Hahn, S., Guarente, L., and Sharp, P. A. (1989)Cell 56, 549561 Buratowski, S., Sopta, M., Greenblatt, J., and Sharp, P. A. (1991)Proc. Natl. Acad. Sci. U. S. A . 88,7509-7513 Carcamo, J., Lobos, S., Merino, A,, Buckbinder, L., Weinmann, R., Natarajan, V., and Reinberg, D. (1989)J. Biol. Chem. 264,7704-7714 Carcamo, J., Maldonado, E., Cortes, P., Ahn, M.-H., Ha, I.-H., Kasai, Y., Flint, J., and Reinberg, D.(1990)Genes & ,Deu. 4,1611-1622 Carcamo, J., Buckbinder, L., and Remberg, D. (1991)Proc. Natl. Acad. Sci. A. as, 8052-8056 Cavallini, B., Huet, J., Plassat,J. L., Sentenac,A., Egly, J.-M., and Chambon, P. (1988)Nature 334,77-80 Chasman, D. I., Lue, N. F., Buchman, A. R., Lapointe, J. W., Lorch, Y., and Kornberg, R. D. (1990)Genes & Deu. 4,503-514 Chen, W., and Struhl,K. (1988)Proc. Natl. Acad. Sci. U. S. A . 85, 2691-2695 Chen, W., and Struhl, K. (1989)EMBO J. 8,261-268 Clark-Adams, C. D., Norris, D., Osley, M. A,, Fassler, J. S., and Winston, F. (1988)Genes & Deu. 2, 150-159 Colgan, J., and Manley, J. M. (1992)Genes & Deu. 6,304-315 Cormack, B. P., and Struhl,K. (1992)Cell 69,685-696 Davison, B. L., Egly, J.-M., Mulvihill, E. R., and Chambon, P. (1983)Nature 301,680-686 Durrin, L. K.. Mann, R. K, Kavne, P. S., and Grunstein, M. (1991)Cell 65, 1023-1031 Dynlacht, B. D., Hoey, T., and Tjian, R. (1991)Cell 66,563-576 Fedor, M. J., Lue, N. F., and Kornberg, R. D. (1988)J. Mol. Biol. 204, 109127 Finley, R. L., Jr., and West, R. W. Jr. (1989)Mol. Cell. Biol. 9, 4282-4290

u.s.

Finley, R. L., Jr., Chen, S., Ma, J., Byrne, P., and West, R. W., Jr. (1990)Mol. Cell. Bid. 10, 5663-5670 Fire, A., Samuels, M., and Sharp, P. A. (1984)J.Biol. Chem. 259, 2509-2516 Flana an, P.M., Kelleher, R. J., Sayre, M. H., Tscbochner, R. D., and Kornberg, R. (1991)Nature 350,436-438 Fong, T. C., and Emerson, B. M. (1992)Genes & Deu. 6,521-532 Gill, G., and Ptashne, M. (1988)Nature 334,721-723 Giniger, E., Varnum, S. M., and Ptashne, M.(1985)Cell 40,767-774 Guarente, L. (1988)Cell 52,303-305 Hahn, S., Buratowski, S., Sharp, P. A., and Guarente, L. (1989)CeU 58,11731181 Han,"., and Grunstein, M. (1988)Cell 55,1137-1145 Han, M., Kim, U.-J., Kayne, P. S., and Grunstein, M. (1988)EMBO J. 7, 2221-2228 Harbury, P. A. B., and Struhl, K. (1990)Mol. Cell. Biol. 9,5298-5304 Hill, D. E., Hope, I. A., Macke, J. P., and Struhl, K. (1986)Science 234, 451457 Homa, F. L., Glorioso, J. C., and Levine, M. (1988)Genes & Deu. 2,40-53 Horikoshi, M., Bertuccioli, C., Takada, R., Wang, J., Yamamoto, T., and Roeder, R. G. (1992)Proc. Natl. A d . Sci. U. S. A . 89, 1060-1064 Johnston, M., and Davis, R. W. (1984)Mol. Cell. Btol. 4, 1440-1448 Ju, Q., Morrow, B. E., and Warner, J. R. (1990)Mol. CeU. Biol. 10,5226-5234 Kelleher, R. J., 111, Flanagan, P. M., and Kornberg, R. D. (1990)Cell 61,12091215 Koleske, A. J., Buratowski, S., Nonet, M., and Young, R. A. (1992)Cell 69, 883-894 Laybourn, P. J., and Kadonaga, J. T. (1991)Science 254,238-245 Lieberman , P. M., and Berk, A. J. (1991)Genes & Deu. 5, 2441-2454 Maldonado, E., Ha, I., Cortes, P., Weis, L., and Reinberg, D. (1990)Mol. Cell. Biol. 10,6335-6347 McCormick, A., Brady, H., Fukushima, J., and Karin,M. (1991)Genes & Deu. 5, 1490-1503 Meisterernst, M., and Roeder, R. G. (1991)Cell 67,557-567 Meisterernst, M., Roy, A.L., Lieu, H. M., and Roeder, R. G. (1991)Cell 66, 981-993 Morrow, B. E., Johnson, S. P., and Warner, J. R. (1989)J. BioL Chem. 264, 9061-9068 Nakajima, N., Horikoshi, M., and Roeder, R. G. (1988)Mol. Cell. Biol.8,40284040 Oliphant, A. R., Nussbaum, A. L., and Struhl, K. (1986)Gene (Amst.) 44,177-

8.

ll)d

Parker C. S. and Topol J. (1984)Cell 36,357-369 Pugh, B. F., And Tjian, R. (1990)Cell 61, 1187-1197 Pugh, B. F., and Tjian, R. (1991)Genes & Deu. 5,1935-1945 Reach, M., Xu, L.-X., and Young, C. S. H. (1991)EMBO J. 10,3439-3446 Reinberg, D., Horikoshi, M., and Roeder, R. G. (1987)J. Biol. Chem. 262, 3322-3330 Rosenberg, A. H., Lade, B. N., Chiu, D., Lin, S., Dunn, J. J., and Studier, F. W. (1987)Gene (Amst.) 56,126-135 Safer, B., Reinberg, D., Jacob, W. F., Maldonado, E., Carcamo, J., Garfinkel, S., and Cohen, R. (1991)J. Biol. Chem. 266, 10989-10994 Sambrook J. Fritsch E. F. andManiatis, T. (1989)Molecular CloninA *rat& 'MQ~uu~,' 2nd id., Cold Spring Harbor Laboratory Press, t o l d Sprmg Harbor,NY Sawadogo, M., and Roeder, R. G. (1985)Cell 43,165-175 Schultz, M. C., Reeder, R. H., and Habn, S. (1992)Cell 69,697-702 Sharp, P.A. (1992)Cell 68,819-821 Simon, M. C., Fisch, T. M., Benecke, J. B., Nevins, J. R., and Heintz,N. (1988) Cell 52, 723-729 Smale, S. T., Schmidt, M. C., Berk, A. J., and Baltimore, D. (1990)Proc. Natl. Acad. Sci. U. S. A . 87.4509-4513 Smale, S. T., and Baltimore, D. (1989)Cell 57, 103-113 Stone, N., Flores, O., and Reinber , D (1991)Pharm. Technol. 15,36-46 Stringer K. F. In les C J , and treenblatt, J. (1990)Nature 345, 783-786 Strubin.".. ahd Sftruhl. (1992)Cell 68.721-730 Struhl, K. (i982)Nature 300,284-287 ' Struhl, K. (1985)Nucleic Acids Res. 13,8587-8601 Struhl, K. (1986)Mol. Cel. Biol. 6,3847-3853 Struhl, K. (1989)Annu. Reu. Biochem. 58,1051-1077 Tabor, S. (1990)in Current Protocols in Mokclllor Biology (Ausubel, F. A. Brent, R., Kin ston, R E , Moore, D. D., Seldman,,J. G., Smith J. A., and Struhl, K., edsy pp. 16:2.i-16.2.11,Greene Publlshmg and Wdeiy Intersclenre --.-- , -NPW .- .. Ynrk - - -.. Tanese, N., Pugh, B. F., and Tjian R. (1991)Genes & Deu. 5,2212-2224 Taylor, I. A., and Kingston, R. E. (1990)Mol. Cell. Biol. 10,165-175 Timmers. H. T. M.. and Sham. P. A. (1991)Genes & Deu. 5. 1946-1956 Van Dyke, M. W., Roeder, R.'G., and Sawadogo, M. (1988)Science 241: 13351338 Wang, H2-Nicholson, P. R., and Stillman, D. J. (1990)Mol. Cell. Biol. 10,

K: